[Serve] Update AIR Examples to use new API, add linked guide (ray-pro…

…ject#27733)

Signed-off-by: Stefan van der Kleij <[email protected]>

doc/source/_toc.yml

@@ -172,6 +172,7 @@ parts:
  - file: serve/handling-dependencies
  - file: serve/http-guide
  - file: serve/handle-guide
+ - file: serve/ray-air-serve-guide
  - file: serve/ml-models
  - file: serve/batching-guide
  - file: serve/model_composition

doc/source/ray-air/doc_code/air_key_concepts.py

@@ -126,14 +126,11 @@ async def adapter(request: Request):
  return pd.DataFrame.from_dict(content)
 
 
-serve.start(detached=True)
-deployment = PredictorDeployment.options(name="XGBoostService")
-
-deployment.deploy(
- XGBoostPredictor, result.checkpoint, batching_params=False, http_adapter=adapter
+serve.run(
+ PredictorDeployment.options(name="XGBoostService").bind(
+ XGBoostPredictor, result.checkpoint, batching_params=False, http_adapter=adapter
+ )
 )
-
-print(deployment.url)
 # __air_deploy_end__
 
 # __air_inference_start__
@@ -142,6 +139,6 @@ async def adapter(request: Request):
 sample_input = test_dataset.take(1)
 sample_input = dict(sample_input[0])
 
-output = requests.post(deployment.url, json=[sample_input]).json()
+output = requests.post("http:https://localhost:8000/", json=[sample_input]).json()
 print(output)
 # __air_inference_end__

doc/source/ray-air/examples/serving_guide.ipynb

@@ -227,12 +227,10 @@
  "from ray.serve import PredictorDeployment\n",
  "from ray.serve.http_adapters import pandas_read_json\n",
  "\n",
- "\n",
- "serve.start(detached=True)\n",
- "deployment = PredictorDeployment.options(name=\"XGBoostService\")\n",
- "\n",
- "deployment.deploy(\n",
- " XGBoostPredictor, result.checkpoint, http_adapter=pandas_read_json\n",
+ "serve.run(\n",
+ " PredictorDeployment.options(name=\"XGBoostService\").bind(\n",
+ " XGBoostPredictor, result.checkpoint, http_adapter=pandas_read_json\n",
+ " )\n",
  ")"
  ]
  },
@@ -272,7 +270,7 @@
  "sample_input = test_dataset.take(1)\n",
  "sample_input = dict(sample_input[0])\n",
  "\n",
- "output = requests.post(deployment.url, json=[sample_input]).json()\n",
+ "output = requests.post(\"http:https://localhost:8000/\", json=[sample_input]).json()\n",
  "print(output)"
  ]
  },
@@ -379,13 +377,12 @@
  "from ray import serve\n",
  "from ray.serve import PredictorDeployment\n",
  "\n",
- "# Launch a Ray cluster running Ray Serve\n",
- "serve.start()\n",
- "\n",
  "# Deploy the model behind HTTP endpoint\n",
- "PredictorDeployment.options(name=\"Adder\").deploy(\n",
- " predictor_cls=AdderPredictor,\n",
- " checkpoint=local_checkpoint\n",
+ "serve.run(\n",
+ " PredictorDeployment.options(name=\"Adder\").bind(\n",
+ " predictor_cls=AdderPredictor,\n",
+ " checkpoint=local_checkpoint\n",
+ " )\n",
  ")"
  ]
  },
@@ -422,7 +419,7 @@
  ],
  "source": [
  "import requests\n",
- "resp = requests.post(\"http:https://localhost:8000/Adder/\", json={\"array\": [40]})\n",
+ "resp = requests.post(\"http:https://localhost:8000/\", json={\"array\": [40]})\n",
  "resp.raise_for_status()\n",
  "resp.json()"
  ]
@@ -504,10 +501,12 @@
  "source": [
  "from ray.serve.http_adapters import pandas_read_json\n",
  "\n",
- "PredictorDeployment.options(name=\"DataFramePredictor\").deploy(\n",
- " predictor_cls=DataFramePredictor,\n",
- " checkpoint=local_checkpoint,\n",
- " http_adapter=pandas_read_json\n",
+ "serve.run(\n",
+ " PredictorDeployment.options(name=\"DataFramePredictor\").bind(\n",
+ " predictor_cls=DataFramePredictor,\n",
+ " checkpoint=local_checkpoint,\n",
+ " http_adapter=pandas_read_json\n",
+ " )\n",
  ")"
  ]
  },
@@ -544,7 +543,7 @@
  ],
  "source": [
  "resp = requests.post(\n",
- " \"http:https://localhost:8000/DataFramePredictor/\",\n",
+ " \"http:https://localhost:8000/\",\n",
  " json=[{\"base\": 1, \"multiplier\": 2}, {\"base\": 3, \"multiplier\": 4}],\n",
  " params={\"orient\": \"records\"},\n",
  ")\n",
@@ -599,10 +598,12 @@
  "source": [
  "from ray.serve.http_adapters import pandas_read_json\n",
  "\n",
- "PredictorDeployment.options(name=\"DataFramePredictor\").deploy(\n",
- " predictor_cls=DataFramePredictor,\n",
- " checkpoint=local_checkpoint,\n",
- " http_adapter=our_own_http_adapter\n",
+ "serve.run(\n",
+ " PredictorDeployment.options(name=\"DataFramePredictor\").bind(\n",
+ " predictor_cls=DataFramePredictor,\n",
+ " checkpoint=local_checkpoint,\n",
+ " http_adapter=our_own_http_adapter\n",
+ " )\n",
  ")"
  ]
  },
@@ -641,7 +642,7 @@
  ],
  "source": [
  "resp = requests.post(\n",
- " \"http:https://localhost:8000/DataFramePredictor/\",\n",
+ " \"http:https://localhost:8000/\",\n",
  " params={\"base\": 10, \"multiplier\": 4}\n",
  ")\n",
  "resp.raise_for_status()\n",
@@ -692,9 +693,11 @@
  }
  ],
  "source": [
- "PredictorDeployment.options(name=\"BatchSizePredictor\").deploy(\n",
- " predictor_cls=BatchSizePredictor,\n",
- " checkpoint=local_checkpoint,\n",
+ "serve.run(\n",
+ " PredictorDeployment.options(name=\"BatchSizePredictor\").bind(\n",
+ " predictor_cls=BatchSizePredictor,\n",
+ " checkpoint=local_checkpoint,\n",
+ " )\n",
  ")"
  ]
  },
@@ -766,7 +769,7 @@
  " futs = [\n",
  " pool.submit(\n",
  " requests.post,\n",
- " \"http:https://localhost:8000/BatchSizePredictor/\",\n",
+ " \"http:https://localhost:8000/\",\n",
  " json={\"array\": [i]},\n",
  " )\n",
  " for i in range(10)\n",
@@ -805,10 +808,12 @@
  }
  ],
  "source": [
- "PredictorDeployment.options(name=\"BatchSizePredictor\").deploy(\n",
- " predictor_cls=BatchSizePredictor,\n",
- " checkpoint=local_checkpoint,\n",
- " batching_params={\"max_batch_size\": 10, \"batch_wait_timeout_s\": 5}\n",
+ "serve.run(\n",
+ " PredictorDeployment.options(name=\"BatchSizePredictor\").bind(\n",
+ " predictor_cls=BatchSizePredictor,\n",
+ " checkpoint=local_checkpoint,\n",
+ " batching_params={\"max_batch_size\": 10, \"batch_wait_timeout_s\": 5}\n",
+ " )\n",
  ")"
  ]
  },
@@ -874,7 +879,7 @@
  " futs = [\n",
  " pool.submit(\n",
  " requests.post,\n",
- " \"http:https://localhost:8000/BatchSizePredictor/\",\n",
+ " \"http:https://localhost:8000/\",\n",
  " json={\"array\": [i]},\n",
  " )\n",
  " for i in range(10)\n",
@@ -902,7 +907,7 @@
  "hash": "3d292ac03404f7ef6351ef2ed79a7f4abdf879ab5af2497c4aeba036dba01cfa"
  },
  "kernelspec": {
- "display_name": "Python 3.6.9 ('base')",
+ "display_name": "Python 3 (ipykernel)",
  "language": "python",
  "name": "python3"
  },
@@ -916,10 +921,9 @@
  "name": "python",
  "nbconvert_exporter": "python",
  "pygments_lexer": "ipython3",
- "version": "3.6.9"
- },
- "orig_nbformat": 4
+ "version": "3.8.5"
+ }
  },
  "nbformat": 4,
- "nbformat_minor": 2
+ "nbformat_minor": 4
 }

doc/source/ray-air/examples/tfx_tabular_train_to_serve.ipynb

@@ -841,19 +841,16 @@
  " Returns:\n",
  " serve URL.\n",
  " \"\"\"\n",
- " serve.start(detached=True)\n",
- " deployment = PredictorDeployment.options(name=name)\n",
- " deployment.deploy(\n",
- " TensorflowPredictor,\n",
- " checkpoint,\n",
- " # This is due to a current limitation on Serve that's\n",
- " # being addressed.\n",
- " # TODO(xwjiang): Change to True.\n",
- " batching_params=dict(max_batch_size=2, batch_wait_timeout_s=5),\n",
- " model_definition=model_definition,\n",
- " http_adapter=adapter,\n",
+ " serve.run(\n",
+ " PredictorDeployment.options(name=name).bind(\n",
+ " TensorflowPredictor,\n",
+ " checkpoint,\n",
+ " batching_params=dict(max_batch_size=2, batch_wait_timeout_s=5),\n",
+ " model_definition=model_definition,\n",
+ " http_adapter=adapter,\n",
+ " )\n",
  " )\n",
- " return deployment.url\n"
+ " return f\"http:https://localhost:8000/\""
  ]
  },
  {
@@ -930,69 +927,15 @@
  "name": "stdout",
  "output_type": "stream",
  "text": [
- "request0 prediction: 0.004963837098330259 - label: True\n"
- ]
- },
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "request1 prediction: 6.652726733591408e-05 - label: False\n"
- ]
- },
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "request2 prediction: 0.00018405025184620172 - label: False\n"
- ]
- },
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "request3 prediction: 0.00016512417641934007 - label: False\n"
- ]
- },
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "request4 prediction: 0.00015515758423134685 - label: False\n"
- ]
- },
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "request5 prediction: 5.948602483840659e-05 - label: False\n"
- ]
- },
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "request6 prediction: 9.51739348238334e-05 - label: False\n"
- ]
- },
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "request7 prediction: 3.4787988170137396e-06 - label: False\n"
- ]
- },
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "request8 prediction: 0.00010751552326837555 - label: False\n"
- ]
- },
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
+ "request0 prediction: 0.004963837098330259 - label: True\n",
+ "request1 prediction: 6.652726733591408e-05 - label: False\n",
+ "request2 prediction: 0.00018405025184620172 - label: False\n",
+ "request3 prediction: 0.00016512417641934007 - label: False\n",
+ "request4 prediction: 0.00015515758423134685 - label: False\n",
+ "request5 prediction: 5.948602483840659e-05 - label: False\n",
+ "request6 prediction: 9.51739348238334e-05 - label: False\n",
+ "request7 prediction: 3.4787988170137396e-06 - label: False\n",
+ "request8 prediction: 0.00010751552326837555 - label: False\n",
  "request9 prediction: 0.060329731553792953 - label: True\n"
  ]
  }
@@ -1012,7 +955,7 @@
  "provenance": []
  },
  "kernelspec": {
- "display_name": "Python 3.7.10 ('ray3.7')",
+ "display_name": "Python 3 (ipykernel)",
  "language": "python",
  "name": "python3"
  },
@@ -1026,7 +969,7 @@
  "name": "python",
  "nbconvert_exporter": "python",
  "pygments_lexer": "ipython3",
- "version": "3.7.10"
+ "version": "3.8.5"
  },
  "vscode": {
  "interpreter": {